If an LLM's prompts allow you to upload files, upload a picture and then ask the LLM questions about it. (Both of the pictures in the examples below were produced by Midjourney.)
(This is slightly different but some AI art sites will produce a picture from your prompt, then modify it as you talk to it, giving further instructions.)
Prompt: Please describe this image and suggest a prompt that would produce a similar image.
Gemini 2.0 Flash replied:
Absolutely! Let's break down this image and create a prompt for a similar one.
Image Description:
The image presents a captivating abstract composition with a strong spiral motif. Here's a more detailed breakdown:
Spiral Dominance: The central element is a prominent spiral that draws the eye inward. It's not a perfectly smooth spiral but rather appears to be constructed from segmented, geometric shapes.
Geometric Texture: The entire image is filled with a mosaic-like pattern of small, angular shapes, resembling fractured tiles or facets. These shapes create a rich, textured surface.
Color Palette: The color scheme is varied and vibrant, featuring a range of warm and cool tones. There are blues transitioning to greens, oranges, yellows, and deep reds/purples. The colors are applied in a way that suggests depth and layering, contributing to the overall complexity.
Fractal-like Quality: The segmented nature of the spiral and the textured surface give the impression of a fractal pattern, where similar shapes are repeated at different scales.
Artistic Style: The image has a painterly quality, reminiscent of digital art or a highly detailed mosaic. It feels both modern and somewhat organic, despite the geometric foundation.
Prompt Suggestion:
Here's a prompt that aims to capture the essence of this image, suitable for AI art generators like Midjourney, DALL-E 2, or Stable Diffusion:
"Create an abstract digital painting with a central, segmented spiral composed of small, geometric facets. The spiral should be the focal point, drawing the eye inwards. Use a vibrant color palette with blues, greens, oranges, yellows, and deep reds, blending them smoothly. The image should have a mosaic-like texture, with a fractal-like quality to the repeated geometric shapes. Aim for a painterly feel, suggesting depth and layering."
Optional Additions to the Prompt (for more specific results):
Specify a medium: "Digital art," "Abstract mosaic," "Fractal art."
Mention a style: "Inspired by cubism," "Art Deco influence," "Modern abstract."
Add emotional keywords: "Dynamic," "Intricate," "Mesmerizing," "Energetic."
Specify a color temperature: "Warm and cool color balance," "Predominantly warm tones."
Add a specific aspect ratio: "--ar 16:9" (for widescreen), "--ar 1:1" (for square).
Key Considerations:
Iteration is key: AI art generators often require multiple iterations and prompt adjustments to achieve the desired outcome.
Experiment with keywords: Try different combinations of keywords to see how they affect the generated image.
Use image prompts (if supported): If your AI tool allows it, you can use the original image as a starting point to guide the generation process.
Let me know if you'd like me to refine the prompt further or explore other variations!
Entering the suggested prompt into Midjourney produced the following four graphics.
Prompt: Please describe this image and suggest a prompt that would produce a similar image.
Gemini said it could not work on pictures of people but Perplexity replied:
Okay, here's a description of the image and a prompt to generate a similar one:
The image shows a young woman with dark hair, eyes closed, and face tilted upwards. She is wearing a light-colored, loose-fitting garment. Her hands are open and raised, as if she is releasing or receiving something. A white dove with its wings spread is flying towards her hands. The background is a soft, blurred mix of blues and greens, with some bokeh-like light effects. The overall mood is peaceful, ethereal, and spiritual. The color palette is cool, dominated by blues and greens.
Here's a prompt you could use to generate a similar image, keeping in mind the artistic style and content:
"A young woman with dark, short hair, eyes closed, face tilted up, wearing a flowing light-colored dress, releasing a white dove into a misty, teal-colored forest, ethereal lighting, soft focus, bokeh, spiritual, serene, cinematic, --ar 3:2"
Subject: "A young woman with dark, short hair, eyes closed, face tilted up, wearing a flowing light-colored dress, releasing a white dove" - Specifies the main elements.
Setting: "into a misty, teal-colored forest" - Describes the background.
Lighting & Style: "ethereal lighting, soft focus, bokeh" - Adds to the mood and visual style.
Mood: "spiritual, serene" - Reinforces the desired emotional tone.
Artistic Style: "cinematic" - Suggests a high-quality, visually appealing image.
Aspect Ratio: "--ar 3:2" - Sets the aspect ratio to be similar to the original.
You can adjust this prompt based on the specific AI art generator you are using and the level of detail you want to control. Good luck! I hope you get some great results.
How well did it do? Putting the suggested prompt back into Midjourney produced the following...